signals-scout-general

Pass

Audited by Gen Agent Trust Hub on Jul 21, 2026

Risk Level: SAFEPROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
  • [PROMPT_INJECTION]: The skill processes untrusted data from various PostHog project sources, creating a surface for indirect prompt injection.
  • Ingestion points: Data enters the agent context through tools such as query-logs, execute-sql, read-data-schema, and previous run scratchpads (signals-scout-scratchpad-search).
  • Boundary markers: Absent. The instructions do not define delimiters or specific warnings to ignore instructions embedded within the processed data.
  • Capability inventory: The skill has the ability to emit and edit reports (emit-report, edit-report) and update a persistent internal scratchpad (signals-scout-scratchpad-remember).
  • Sanitization: Absent. There is no evidence of sanitizing, escaping, or validating the external data before it is used to influence report authorship or scratchpad updates.
  • [COMMAND_EXECUTION]: The skill utilizes the execute-sql tool to perform database queries. While intended for data analysis, dynamic SQL execution based on derived insights represents a powerful interface that requires human oversight to ensure queries are not influenced by malicious data patterns found in the logs.
Audit Metadata
Risk Level
SAFE
Analyzed
Jul 21, 2026, 08:12 PM
Security Audit — agent-trust-hub — signals-scout-general